agentleFS
netanelavrnetanelavr/dotcursor/skills/kit-review/SKILL.md

kit-review

Skill6 starsChanged 27 days agoMITView on GitHub

Review a branch or PR - author self-check (changed files only) or reviewer second-pass (architecture + diff). Use before publish or when reviewing someone else's PR. Modes are author or reviewer.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 4 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Not a license to rewrite unrelated code (Boy Scout only when asked)

    line 18
  2. Very specific:

    Not proof that it works (kit-prove) and not blast-radius (kit-check-blast)

    line 19
  3. Very specific:

    Not stripping unused additions, extra instruction lines, or format-only hunks (kit-strip-diff)

    line 20
  4. Very specific:

    Not the feature decision log (kit-decisions) — read it when present.

    line 21
  5. Distinctive:

    If nothing meaningful: say the diff is acceptable.

    line 30
  6. Very specific:

    If this change belongs to a feature that has a decisions.md (docs/features/<feature>/decisions.md, or a file that already sits with that …

    line 34
  7. Distinctive:

    Prefer fewer, higher-value comments

    line 82
  8. Distinctive:

    No generic advice or hypothetical edge cases

    line 83

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Review

Valuable, trustworthy feedback — not a wall of nits.

Before reviewing, read references/review-checklist.md.

Accept kit-review author|reviewer. If omitted: author when the user owns the uncommitted/branch diff; reviewer when they asked for a second pass, a PR URL, or “review this PR”.

What this is NOT

  • Not “comment on every line” — prefer fewer, higher-severity findings
  • Not a license to rewrite unrelated code (Boy Scout only when asked)
  • Not proof that it works (kit-prove) and not blast-radius (kit-check-blast)
  • Not stripping unused additions, extra instruction lines, or format-only hunks (kit-strip-diff)
  • Not the feature decision log (kit-decisions) — read it when present; do not rewrite it here

Scope

ModeScope
authorOnly the provided changes (diff / modified files). Unchanged code only if the change directly breaks it.
reviewerDiff vs default branch plus a light approach check. Prefer fewer findings. Full domain/ADR grill → kit-challenge-design.

If nothing meaningful: say the diff is acceptable.

Feature log

If this change belongs to a feature that has a decisions.md (docs/features/<feature>/decisions.md, or a file that already sits with that feature), read it before findings. Maintaining the log is kit-decisions.

  • Re-proposing a Rejected option without new evidence is a finding.
  • Undoing an Accepted decision without new evidence is a finding.
  • A Landmine the diff steps on is a finding, even when the edit looks small.

Approach

  1. Big picture — intent + system impact
  2. Flows — main paths (happy + failure)
  3. Drill down — checklist P0 → P3
  4. Alternatives — only if they clearly improve simplicity/correctness/reuse
  5. Reviewer-only — light approach check (why this path; simpler option; failure modes). Deep domain/ADR grilling is kit-challenge-design. Skip hypothetical 10× scale unless the change is actually about scale.

Philosophy

  • Impact over certainty
  • Raise issues likely to cause: maintenance burden, fragile behavior, hidden bugs, security/data risk
  • Engineering judgment, not lint echo
  • Flag speculative generality, single-use abstractions, and new modules that duplicate in-repo behavior

Response structure

1. Big Picture Summary

  • Intent / system impact / key trade-offs

2. What / Where

  • Concise change list + files

3. Flows & Logic

  • Per main flow: entry, happy path, failure/edges, side effects

4. Findings (checklist order)

  • Critical (must fix)
  • Improvements
  • Missing tests (critical paths)
  • Boy Scout (optional)

5. Could this be done better?

  • Alternative + why + trade-offs — only if meaningful

6. Ready?

  • author: publish / fix-first
  • reviewer: approve / request changes + the bar that was not met

Output guidelines

  • Clear and direct
  • Prefer fewer, higher-value comments
  • No generic advice or hypothetical edge cases
  • Pair “it works” claims with kit-prove if validation is in scope

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.